Sam Altman told OpenAI‘s board in October 2022 that the company intended to release a language model matching GPT-3 capabilities on consumer hardware before Stability or competitors did. The email, exposed during the Musk v. Altman trial in 2026, stated that having such a model available would discourage other firms from launching similarly powerful systems and make it harder for new projects to secure funding. Altman described these plans as part of an extensive discussion on open source strategy that the board had not yet fully finalised at that time.
This internal memo highlights how early industry leaders viewed local deployment as a strategic barrier to entry rather than just a technical benefit. By aiming to put powerful AI on personal devices, OpenAI sought to control the market dynamics before rivals could establish their own footholds. The strategy relied on the assumption that accessibility would stifle competition, a logic that has since faced scrutiny as many smaller teams have successfully built open alternatives.
- The target was a model comparable to GPT-3 running on standard consumer hardware
- Altman believed this move would reduce funding opportunities for new entrants
- The plan was discussed in October 2022 but remained unconfirmed by the board
